Library / SDK
warp-drive-data/warp-drive avatar
warp-drive-data/warp-drive

WarpDrive: the API surface is tracked in four JSON files at the root

WarpDrive is the lightweight data library for ambitious web apps — universal, typed, reactive, and ready to scale.

3,161 stars1,339 forksTypeScriptMIT

At a glance

What is it?
The data library the Ember team ships for web apps, with a package family split across two scopes and two source roots. What is unusual is how much of the README is generated, and how carefully the compatibility policy is worded.
Who is it for?
WarpDrive is a reasonable choice for a team already inside the Ember ecosystem that wants a typed data layer with server-side rendering support and does not want to re-architect an existing API. Two things to check before committing.
Can I use it commercially?
Yes. MIT is a permissive licence: you can use, modify and sell software built on it, as long as you keep its copyright and licence notices.
Is it still maintained?
Yes. The repository last received commits 2 days ago.
What is it written in?
Mainly TypeScript, according to GitHub's language statistics.

Answers come from the project's GitHub data, last synced on November 13, 2026, and from our analysis. They are not legal advice.

From daily snapshots

Star history

  • warp-drive-data/warp-drive3,161 ★ · +4
Star history
Repository2026-09-242026-11-13
warp-drive-data/warp-drive3.2k3.2k

4 Stars in the last 50 days, measured from the daily GitHub snapshot. Compare with other projects

Editorial analysis

One library, two package scopes, two source roots

The version table lists more than twenty packages, and reading the paths rather than the names explains why. Two source roots are in play, packages/ and warp-drive-packages/, and packages appear under both. Newer ones carry the @warp-drive scope: schema-record, holodock, experiments, ember, diagnostic, core-types, and build-config. Older ones keep the @ember-data scope, including store, model, adapter, serializer, request, rest, json-api, graph, active-record, tracking, and codemods. That split is the visible edge of a rename rather than a clean architecture, and the audience column in the same table flags which packages serve both ecosystems, with a mixture of the two markers on a few of them. Two more entries sit outside the data path proper: an eslint plugin for the project, and a package directory named with a leading dash, which is how the umbrella distribution is kept separate from the individual packages.

The compatibility table is generated and deliberately conservative

The Ember compatibility table is the most carefully specified artefact in the file, and it is not written by hand. Four columns describe each release's relationship to the framework: lockstep is the latest ember-source at the time of release, supported is what the release officially supports, tested is what the project actually ran tests against, and range is the peer dependency range the release states. The surrounding notes are the part to internalise. The library is often compatible with a larger range than was supported at release, the project makes an effort to test against a wider range than it officially claims, and in many cases it works beyond what the table shows. Where possible they try to keep that wider window open. Both tables are generated from a data file inside the internal tooling with the command pnpm sync-readme-tables, and each is fenced by placeholder markers in the markdown, which is how a generated table survives being pasted into a README.

Dropping an unsupported Ember is not a breaking change

One sentence in the compatibility section settles most upgrade arguments. If it becomes necessary to break compatibility with an older, unsupported release, the project says it does not consider that a breaking change. The footnote markers in the table show the cases that are exceptions rather than rules. A special release exists specifically to support Vite builds in applications still on the 4.x line, with its own row and its own narrower set of tested versions. Two rows are marked with warnings rather than plain failures: one for versions that may receive special long-term patches to help people migrate off ModelFragments and onto the 5.x line, and another whose footnote records that special updates have extended support to ember-source 5 and 6. Those three footnotes are where the actual support policy lives, and they are attached to rows rather than described in prose.

Releases arrive in pairs, seconds apart, with names

The tags carry names, which tells you something about the release culture. Version 5.8.2 is titled Welcome To The Jungle and was published on 2026-04-17. Then two tags land on the same morning: 5.9.0, titled Commencing Countdown, at 07:29:18 UTC on 2026-09-05, and 5.9.1, titled Starships Were Meant To Fly, thirty-seven seconds later at 07:29:55. A minor and a patch release in the same minute is a deliberate double publish rather than an accident, and it means the patch was prepared alongside the minor. The branch is ahead of both. The root manifest is private and versioned 5.10.0-alpha.14, so what is on main is an alpha of the next minor, while the newest published tag is 5.9.1. The repository is not archived and the last push is dated 2026-10-03.

Build concurrency is computed by a script

The root manifest shows the shape of the toolchain, and one pattern repeats. Build, prepare, and every lint target funnel into Turbo with a concurrency value taken from a shell expansion rather than a fixed number.

json
    "prepare": "turbo run build:pkg --log-order=stream --concurrency=$(./bin/calculate-jobs);"

So the prepare hook and the build script are the same command, which means installing the repository builds the packages, and the degree of parallelism comes from a script in bin/. The dev start task takes a different approach and hard-codes a concurrency of 23, and the sync task bypasses Turbo entirely, filtering pnpm across the packages, the second packages root, the tests, and the specs directories with --parallel and --if-present so a directory without a sync script is skipped rather than failing. Publishing is not a Turbo task either: the release script points at a TypeScript file under tools/release, so releasing is a script the project owns rather than a package manager command.

The documentation has its own viewer and a link checker

Documentation is a first-class part of the repository rather than a link away. A docs-viewer directory holds a site with its own preview task, run by changing into the directory and starting the package, and its own check task that validates links, which is why the lint entry points for documentation exist separately from the package lint tasks. Alongside it sit guides, a blog directory, and a next-release-post directory, the last being where release announcements are staged, which pairs with the codename-style tag names. Other root files describe the project to machines rather than to people: a context7 configuration file, a CNAME for the documentation domain, and a logos directory. Tool versions are pinned through a mise configuration rather than through a version field, and there are editor and agent directories for Claude, Cursor, and Gemini alongside agent instruction files at the root.

Four export mappings sit beside the lockfile

Four JSON files at the repository root map what the public API surface looks like, and their names describe the comparisons they support: one for the main branch, one for the released package, one for version 5.5, and an enriched variant of the 5.5 mapping. That is the mechanism behind a claim the README does not make out loud, which is that this project takes its export surface seriously enough to diff it. The same care shows up elsewhere. A patches directory at the root implies dependencies are patched rather than forked, an upgrading directory and a modules-migration document cover the move from the older package layout, and four more first-class roots exist for specifications, tests, RFCs, and tools. The RFC link points at the Ember RFC repository filtered to the data label, which places the design discussion outside this repository while the code lives here.

Editorial conclusion

WarpDrive is a reasonable choice for a team already inside the Ember ecosystem that wants a typed data layer with server-side rendering support and does not want to re-architect an existing API. Two things to check before committing. Read the compatibility notes rather than the table alone, since the project states the real support window is wider than the stated one and that dropping an already unsupported Ember version is not treated as breaking. And pick your package deliberately, because several still ship under the older scope while the project is mid-rename.

Frequently asked questions

What is WarpDrive?

A lightweight data library for web apps, built with Ember and described as universal, typed, reactive, and ready to scale. It advertises seamless reactivity in any framework, a small footprint, typed APIs, works with any API, and server-side rendering support.

Which Ember versions does WarpDrive support?

The compatibility table tracks four things per release: lockstep, the ember-source version current at release time, supported, the officially supported versions, tested, the versions actually tested against, and range, the stated peer dependency range. The project says the real support window is often wider, and that breaking an already unsupported release is not considered a breaking change.

Why are there both @warp-drive and @ember-data packages?

The newer packages carry the @warp-drive scope while older ones keep @ember-data, and they live under two source roots, packages/ and warp-drive-packages/. An audience column in the generated versions table marks which packages serve both ecosystems.

How is the WarpDrive README kept up to date?

Both the Ember compatibility table and the Big List of Versions are generated with the command pnpm sync-readme-tables from a data file inside the repository's internal tooling, and each table sits between placeholder markers in the markdown so a generated block can be replaced safely.

Official sources

  1. License: MIT
  2. Project website
  3. README
  4. Releases
  5. warp-drive-data/warp-drive on GitHub
Add this badge to your README

If you maintain this project, the badge below links readers to this analysis and shows its maintenance status from the daily GitHub snapshot. Paste the markdown into your README; add ?metric=license or ?metric=stars to the image URL for a different field.

Add this badge to your README

markdown
[![Hysen Labs](https://hysenlabs.com/badge/warp-drive-data-warp-drive.svg)](https://hysenlabs.com/projects/warp-drive-data-warp-drive)

Star history chart for your README

markdown
[![Star history](https://hysenlabs.com/chart/warp-drive-data-warp-drive.svg)](https://hysenlabs.com/projects/warp-drive-data-warp-drive)

Was this useful?